10-2

effects or minimally significant effects must be cautious, in view of the
small number of animals used. Only rhodium and palladium (tested in mice
only) showed any signs of carcinogenicity, but as Schroeder and Michener
(1971a) stated, "The results were at a minimally significant level of
confidence." Further studies are needed to confirm these findings.
Schroeder also reported that selenate, but not selenite, increased the
incidence of spontaneous malignant mammary and subcutaneous tumors in
rats after lifetime exposure (11 in 75 controls vs 20 in 73 selenate-fed
animals). These results were not confirmed in similar studies in mice.
(The effects of selenium on carcinogenesis are discussed in further
detail below.) None of the remaining elements examined increased tumor
incidence. A significant reduction in tumor incidence was observed in
mice fed arsenic and cadmium and in mice and rats fed lead.

SELENIUM

Signs of chronic selenium toxicity in animals have been recognized
for almost 700 years, but selenium was not identified as the responsible
agent until the 1930fs. Twenty years later, the economic importance of.
selenosis and selenium deficiency for animal producers became apparent.
This discovery stimulated the mapping of selenium distribution in the
soils, forages, and tissues of humans in several continents. Extreme
differences of exposure were delineated, even within individual countries.
This knowledge enabled investigators to make epidemiological correlations
of diseases, including cancer, in humans and animals and to conduct lab-
oratory experiments to test the resulting hypotheses (National Academy of
Sciences, 1971).

Epidemiological Evidence

Selenium has been reported as having a possible protective effect
against cancer. Shamberger and colleagues correlated selenium levels
in forage crops (grouped into high, medium, and low categories) with
cancer mortality by state in the United States (Shamberger and Frost,
1969; Shamberger and Willis, 1971; Shamberger £t al., 1976). They
found an inverse relationship in both males and females, especially
for cancers of the gastrointestinal and genitourinary jtracts. In other
studies, Schrauzer and coworkers correlated per capita intake with cancer
mortality rates in more than 20 countries (Schrauzer, 1976; Schrauzer £t
al* y 1977a,b). The consumption estimates were based on international
food disappearance data for major food sources (e.g., cereals, meat, and
seafoods) to which the investigators attributed plausible mean selenium
values. They found an inverse relationship between selenium intake and
leukemia as well as with cancers of the colon, rectum, pancreas, breast,
ovary, prostate, bladder, lung (males), and skin. Using pooled blood
samples from healthy donors in 19 U.S. states and 22 countries, they also
correlated blood levels of selenium with corresponding cancer mortality
rates. They found significant inverse relationships for most of these
same sites.